13

We are constructing a new building, and our CIO is pushing for a solution where he doesn't have to spend the money on copper runs to all the user stations and wants to use wireless only for this purpose, especially since wireless will already be part of the deployment.

Servers, the data center, services deemed as critical, and other non-wireless devices (copier/scanner/printers for instance) would still get a copper connection.

Clearly installing copper cabling and the switches to support all the cabling can be expensive, but this seems like a major change.

How viable is it to provide only wireless to user stations?

1
  • With the uncertainties of the upcoming rollout of LTE-U, the telcos' grab to steal and commercialize the Wi-Fi spectrum, it may be time to start thinking about what happens if it does severely degrade Wi-Fi. It will have some impact on Wi-Fi, but nobody yet knows how much. The ostrich approach does nothing for you, but how many, and which, resources do you put into it? it could certainly be crippling if there is a severe impact. – Ron Maupin Oct 11 '15 at 1:58
17

With the arrival of 802.11n and with 802.11ac almost here, this can be a viable option. However the expectation of cost savings may not be accurate as you may have additional initial costs (for instance hiring a wireless consultant to make sure things go right) or even in ongoing costs (do you need new staff?). You will carefully have to consider your path forward and how this impacts your users and IT staff.

That being said, there are a number of things to keep in mind when considering not deploying a traditional 802.3 Ethernet network (the larger the wireless network and number of client devices, the more these are true):

  • 802.11 is somewhat still in the same place where 802.3 was in it's infancy. There are device inter-operability issues that exist based on the way that some vendors implement the standards. Your IT staff may need to spend additional time testing new hardware or software.
  • 802.11 is newer, more complicated, and changing much faster than 802.3, so the development and quality assurance processes by wireless vendors (infrastructure and client devices) is not always what it should be.
  • You will want to make sure that you properly plan the wireless deployment considering a number of factors including at least both coverage and capacity planning.
  • 802.11 can be more difficult to configure, optimize, and maintain properly than a traditional 802.3 network. Having someone on staff or hiring a consultant who are well versed in 802.11 and RF will be very helpful.
  • Wireless is not as reliable as wired since it is more easily subject to outside sources of interference.
  • If you haven't already, you would want to invest in some tools to help your IT staf maintain and troubleshoot the wireless network, including at least a basic 2.4GHz/5GHz spectrum analyser and a wireless capture device (without special hardware/drivers, Windows cannot capture wireless traffic properly; Linux can with the right hardware).
  • Problems in 802.11 can be more difficult to troubleshoot, so problem resolution can take longer at times.
  • There may be additional security concerns and costs with an 802.11 network. At the very least you will want to consider running WPA2-Enterprise security if you don't have it already. VPN for wireless clients is also an option (with or without WPA2).

I can probably go on, but I hope I have illustrated that while this is a viable option, you need to make sure you understand what this change entails if you wish to do it successfully.

2
  • 3
    YL, good advice. For the OP, depending on the number of clients to be served, the cost of entry on an Enterprise-level Unified Wireless Network may be far greater than just going the wired route. By the sound of your CIO trying to go the cheap route, you' can expect far more trouble and complaints with the wireless system if you go that way and skimp on your deployment or cut corners. Do not even consider going wireless without providing 5GHz for your users as the 2.4GHz spectrum is too crowded, fewer channels, and usually has more interference. – generalnetworkerror May 26 '13 at 3:14
  • +1 on the reliability and troubleshooting issues, especially when there is corner cutting on deployments – tegbains May 26 '13 at 5:28
3

What about voice for each station? Whilst it is acceptable these days to perhaps use wireless for all your data needs for desktops it is still the case IMHO that you should be using wired for voice and if you are going to run copper for the voice then why not just use copper for the desktops as well...

3
  • This seems more like a question than an answer, but to address, if you are talking about 500 users and wired voice (so leave out the labor since it wouldn't be too much different), you are still talking of a significant cost in copper, parts, patch cables and switchports. Ongoing costs for maintenance and power for the switchports. Additionally, wireless VoIP is a growing option and has many benefits. I know several companies that have moved to dual function phones (cellular/SIP) that have largely done away with a traditional phone system. – YLearn May 25 '13 at 21:00
  • 1
    it is still the case IMHO that you should be using wired for voice <- you haven't provided any reason for that. – Lie Ryan May 26 '13 at 7:38
  • 4
    Wireless, no matter which way you look at it is a half duplex medium and no matter how much optimisations you put into the AP's to give voice a priority it is never going to match that of standard copper runs. – David Rothera May 26 '13 at 10:49
3

Going "All Wireless" can be very viable in the right environment and with proper planning. One of the key concepts that I see engineers who have not spent a good bit of time working on WiFi networks make is to forget that wireless networks that uses FCC unlicensed spectrum, such as 802.11n or 802.11ac operate on a shared, uncontrolled physical medium. A successful deployment will keep in mind the physical environment as well as the expected network utilization, both factors in managing external (outside your building or floor) and internal, self induced interference. I have seen companies with minimal network traffic requirements go all wireless in even noisy open environments while others with high performance requirements required a great deal of expensive time and effort to tweak and tune the AP (Access Point) density, configuration and placement. I have seen companies with a lot of media heavy users in tight quarters above the ground floor of an all glass walled office building in down town San Francisco that eventually went all wireless, it is just took a lot of work.

Ask yourself a few questions:

  1. What kind and how much traffic should I expect on the WiFi network?
  2. How dense will the client environment be? (# of users per AP can have a significant difference on performance in noisy environments)
  3. What are the walls of the building made of? A Drywall wall might only cause 3dB of signal loss, while poured cement walls might be over 20dB of loss.
  4. What are your neighbors doing? There are plenty of simple, free tools that you can use to see what the wireless environment outside the building/office is like.

A typical CAT6 cable run will cost between $150-$300 based on your location, do the math and it is easy to see that a well engineered WiFi network can save you money and, if you plan carefully, keep your users happy.

2

In order to get a good answer it would help to know how large an organization you are planning for, both in terms of area, isolation from other organizations, the number of users to support and intensity of those users' computer usage).

I've been at a few companies that went all-wireless for computer access; it worked because they were relatively small, and there were free wifi channels. Larger installations are more likely to suffer from congestion at peak usage times.

And there is always the issue of a neighboring company setting up stations "on your channels", resulting in much lower throughput.

There is the issue of which bands you are going to use: 802.11b/g or 802.11a. There are only three clean channels in 802.11b/g, there are many more in 802.11a, as well as supporting higher native throughput and smaller cell sizes.

I usually recommend against a wireless only installation; given how critical internet access is for most of today's workers it doesn't take very many wireless outages to justify the extra $300/station capital costs for hard-wiring a station.

Finally, have you done a site survey to see how much competing wireless usage there already is in your location?

2
  • You do raise some very good considerations about designing the wireless properly, however I don't think the design directly relates to the core question of is the approach a viable one. While I wouldn't normally recommend this approach either, money sensitive organizations are going to raise this question more and more frequently. – YLearn May 25 '13 at 22:44
  • 1
    My point was: small organizations with light usage and not much interference will do fine. But without more details its hard to tell if you fit that profile. – Neil Katin May 26 '13 at 1:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.